From a user perspective, the network file server is often both the most forgotten piece of IT Infrastructure, and the most sorely missed when it is not there.
You alternatively have the ultimate bad guy if the server is down (or worse, is missing data) and the complete hero who manages to get the system back up and running.
You might even get some sympathy for a day or so if you had to pull an all-nighter to make sure everything was back up and running.
Companies depend on their networks, and the servers are a crucial part of the network. If there is no file server there are no files to work on and no printers to print to.
If the server is also the main authentication server for the company or department, users may not be able to access other essential services including e-mail and the Web proxy server.
For this review, we received mostly systems based on an Intel platform of Intel processors and Intel chipsets.
The notable exceptions were the budget systems from Xenon and Hallmark, which ran with VIA chipsets and the Tachyon 1600MP from Emagen which was the only Athlon based machine we received.
It contained a pair of AMD 1600+ processors (running at 1.4GHz) on a Tyan Tiger motherboard based on AMD's 760MP chipset.
One interesting feature we found in the Intel boards was the ability to push the Esc key during the boot sequence to select the boot device.
This means the system can be left to boot directly from the hard disk, but if you need to boot from other media for an update, you can do it quickly, and without the danger of leaving the system set to boot from a floppy, which can of course be disastrous if you leave the floppy behind when you finish.
This article looks at two scenarios, servers for 100 users, and servers for 500 users.
Which OS?
One of the biggest questions facing most companies is which operating system to choose for their servers. Many times this decision is made by the local sysadmin, based on which OS they are most familiar with, rather than necessarily which OS is better for the company in this situation. There are various things to consider when choosing which operating system is right for your situation:
- What is it doing now?
- What will it be asked to do in the future?
- How will it interoperate with the other systems we are running?
- How much does it cost to purchase?
- How much will it cost to upgrade?
- Can we support it?
- What about uptime?
- What about security?
Getting it right
Now is not the time to start an OS argument, and there was lots of useful information in last month's discussion "OS Wars", so we will simply comment on a few items.
We know you can build a rock solid Windows server that will run for months without unscheduled downtime.
However, it tends to be much easier to build a simple, more secure Linux server that will give better uptime. Netware servers too, have a history of massively long uptimes.
Cost is a very real issue, for the initial purchase price of the OS to some extent, but even more so for the cost of user licenses.
A Windows 2000 server with 100 client access licenses (CALs) costs around AU$7400 (including GST), while 500 CALs costs more like AU$31,600.
Compare this with may of the versions of Linux you can buy for between AU$30 and AU$1000, or even download for free-with unlimited user licensing and free upgrades ("free" doesn't include the costs of downloading, of course).
That was a pretty uninformative story!
Talks about perforamnce etc, prices, etc, without actually giving these figures for every machine.
Even in each of the two categories covered, it was apples to oranges.
hmm...